Instructions to use markab/rpn-gpt-lightning-MLX-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use markab/rpn-gpt-lightning-MLX-4bit with MLX:

# Make sure mlx-lm is installed
# pip install --upgrade mlx-lm

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("markab/rpn-gpt-lightning-MLX-4bit")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

text = generate(model, tokenizer, prompt=prompt, verbose=True)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use markab/rpn-gpt-lightning-MLX-4bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "markab/rpn-gpt-lightning-MLX-4bit"

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "markab/rpn-gpt-lightning-MLX-4bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use markab/rpn-gpt-lightning-MLX-4bit with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "markab/rpn-gpt-lightning-MLX-4bit"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "markab/rpn-gpt-lightning-MLX-4bit"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "markab/rpn-gpt-lightning-MLX-4bit",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use markab/rpn-gpt-lightning-MLX-4bit with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "markab/rpn-gpt-lightning-MLX-4bit"

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default markab/rpn-gpt-lightning-MLX-4bit

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use markab/rpn-gpt-lightning-MLX-4bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "markab/rpn-gpt-lightning-MLX-4bit"

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "markab/rpn-gpt-lightning-MLX-4bit" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

rpn-gpt-lightning-MLX-4bit

rpn-gpt-lightning for Apple silicon, quantized with mlx-lm 0.32.2 (mlx_lm.convert -q --q-bits 4 --q-group-size 64, 4.503 bits per weight, 17 GB): NVIDIA Nemotron 3.5 Lightning (30B, 3B active) fine-tuned to drive an exact Forth machine. The main model card has the results and the full story; code: github.com/markab21/rpn-gpt.

The model only writes the Forth. The machine that runs it is rpn_chat.py, a small client loop.

pip install mlx-lm==0.32.2
mlx_lm.server --model markab/rpn-gpt-lightning-MLX-4bit --port 8090
git clone https://github.com/markab21/rpn-gpt && cd rpn-gpt && uv sync
uv run python rpn_chat.py "Calculate the remainder when 82225043186179281 is divided by 6469193." \
  --server mlx --base-url http://localhost:8090 --model default_model

Tested on an M4 Max (128 GB) with mlx 0.32.4 / mlx-lm 0.32.2: 3 of 3 hard demo problems (demo10 numbers 6, 3, 9) correct through rpn_chat.py --server mlx, 240 s in all, about 62 tokens/s. Accuracy not measured. mlx_lm.server 0.32.2 can stop generating on an exact prompt-cache hit in long sessions; training/mlx_local.sh patches it.

License

NVIDIA Nemotron 3.5 Lightning is under the OpenMDW License Agreement, version 1.1: full text and NVIDIA's notice in LICENSE. Our parts (the fine-tune, the card, the code) are MIT, Copyright (c) 2026 Mark Berry and rpn-gpt contributors (LICENSE-MIT). Please credit rpn-gpt with a link to github.com/markab21/rpn-gpt.

Downloads last month: -

Safetensors

Model size

32B params

Tensor type

U32

BF16

F32

MLX

Hardware compatibility

4-bit

Model tree for markab/rpn-gpt-lightning-MLX-4bit

Base model

nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16

Finetuned

markab/rpn-gpt-lightning

Quantized

(4)

this model

Dataset used to train markab/rpn-gpt-lightning-MLX-4bit

Collection including markab/rpn-gpt-lightning-MLX-4bit

rpn-gpt

Collection

An exact Forth machine in Nemotron 3.5 Lightning: merged model, LoRA, NVFP4, GGUF (Ollama), MLX, the kernel in the weights, and the data. • 8 items • Updated about 13 hours ago